Key Takeaways
- Wells Fargo maintains bullish stance on NVDA approaching GTC 2026, referencing historical outperformance versus SOX index ranging from 12% to 45% in three-month periods post-conference
- Bank of America maintains Buy rating with $300 price target, highlighting NVDA’s current valuation at approximately 17x forward PE—a historical bottom
- Second-generation co-packaged optic switch anticipated to be announced, along with progress reports on Feynman GPU series and Kyber NVL576 rack systems
- The rise of agentic AI workflows is fueling renewed CPU demand — Nvidia’s Vera CPU has entered production with deployment at Meta facilities, broader rollout scheduled for 2027
- Industry-wide CPU supply constraints emerging, with both AMD and Intel reporting lead times extending to six months and price increases exceeding 10%
As Nvidia (NVDA) prepares for its flagship GTC conference scheduled for March 16–19 in San Jose, California, Wall Street analysts are positioning the event as a potential inflection point for both the chipmaker and the semiconductor industry at large.
Aaron Rakers and his team at Wells Fargo have expressed a positive outlook, stating they are “NVDA buyers ahead of the event.” Their analysis highlights a consistent trend of robust stock appreciation in the three-month window following previous GTC gatherings, with NVDA typically outperforming the SOX semiconductor benchmark by approximately 30% on average, spanning a range from +12% to +45%.
Bank of America’s Vivek Arya has maintained his Buy rating alongside a $300 price objective. His commentary emphasizes that shares are currently valued at roughly 17x forward earnings — representing a historical valuation floor — following the successful Blackwell product launch that has generated an estimated $500 billion in aggregate revenue.
CEO Jensen Huang is scheduled to present his keynote presentation at 2 p.m. ET on Monday. He will additionally lead an industry discussion panel on Wednesday afternoon. Featured participants in main stage presentations include OpenAI, Google DeepMind, Meta, Microsoft, and Tesla.
Regarding product announcements, Nvidia is anticipated to introduce its next-generation co-packaged optic switch, incorporating Taiwan Semiconductor’s co-packaged optic capabilities. Mass production is projected to begin in 2027, targeting approximately 80,000 units. Additional updates on the Feynman GPU family and the Kyber NVL576 rack architecture are also expected.
Wells Fargo anticipates Nvidia will refresh its pipeline projections, potentially elevating its cumulative revenue forecast from $500 billion to north of $600 billion through 2026. Rakers additionally highlighted the possibility of updated estimates for global AI infrastructure expenditure, currently projected at $3–$4 trillion annually by 2030.
CPUs Emerge as Critical Component
While GPUs remain central to Nvidia’s narrative, a significant transition is occurring beneath the surface. Agentic AI — workflow-orchestrating AI that coordinates multiple autonomous agents — demands a distinct computational architecture compared to conventional AI inference workloads. This evolution is driving unprecedented demand for central processing units.
Dion Harris, Nvidia’s head of AI infrastructure, explained to CNBC this week that “CPUs are becoming the bottleneck in terms of growing out this AI and agentic workflow.” The company’s Vera CPU has entered production phase and is currently operational within Meta data centers following a multiyear agreement announced in February. Nvidia intends to broaden that deployment in 2027.
Thousands of independent Nvidia CPUs are currently operational at the Texas Advanced Computing Center and Los Alamos National Lab. Bank of America forecasts the CPU market could expand dramatically, growing from $27 billion in 2025 to $60 billion by 2030.
At GTC, Nvidia is anticipated to display a CPU-exclusive rack configuration on the exhibition floor — demonstrating the company’s strategic commitment to standalone CPU infrastructure.
Supply Constraints Intensify
The wider CPU marketplace faces mounting pressure. AMD and Intel have both issued warnings regarding supply limitations, with fulfillment timelines extending to six months and pricing climbing over 10%, according to Reuters reporting.
Forrest Norrod, AMD’s data center division head, told CNBC that demand surges during the previous six to nine months have been “unprecedented.” Intel indicated inventory levels will reach their nadir this quarter, though the company anticipates supply conditions will improve throughout Q2 2026.
Currently, Nvidia reports it has not experienced significant CPU shipment disruptions. Harris indicated the company’s supply chain infrastructure has successfully accommodated demand, partially because the majority of its CPUs are delivered integrated within complete rack-scale configurations alongside GPUs.
Mercury Research data shows Nvidia captured a 6.2% segment of the server CPU market during Q4 2025, trailing Intel at 60% and AMD at 24.3%. Additional stocks positioned to respond to GTC developments include AMD, Taiwan Semiconductor, Broadcom (AVGO), Intel, and Marvell (MRVL).
